Import Contacts
Contacts
Start an Import
Start an async import of the stashed CSV with a confirmed mapping.
Inserts a contact_imports job row and starts ContactImportWorkflow,
returning 202 { import_id } immediately. The frontend polls
GET /contacts/imports/{import_id} for progress + results.
POST
Import Contacts
Third step of the CSV import flow. Inserts a
contact_imports job row (status: "pending") and starts ContactImportWorkflow on Temporal, then returns immediately — the actual parsing/inserting happens asynchronously. If the workflow fails to start, the job row is patched to status: "failed" synchronously and the request itself still returns 502.
Rows are matched against mapping (raw CSV header → canonical field, same keys as fields[].key from Preview; "" means “don’t import this column”). A row needs at least one identity field after mapping to be imported; a row is skipped if its email already exists in the org, or repeats an email already used earlier in the same file. Type-checking is lenient — an invalid cell (e.g. a malformed phone number) is dropped and the row still imports as a warning, not a hard failure.
Beyond the 15 MB file-size cap shared by all three steps, there’s a second limit that’s easy to miss: at most 5,000 rows are imported per file. A CSV can sit well under 15 MB and still blow past it. Rows after the 5,000th aren’t a hard failure — each is skipped individually with the reason exceeds the 5,000-row limit, counted in skipped on Get import status and annotated per-row in that job’s result CSV, while the job itself still finishes as completed. Split larger files into separate import jobs.
Auth
Requires a CRM manage scope and an active organization on the token. Any*:manage scope qualifies — in practice contacts:manage, deals:manage, companies:manage, or activities:manage.
Response
202 Accepted:
Errors
Authorizations
Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
Body
application/json
Start an async import of a stashed CSV with a confirmed column mapping.
mapping is raw-CSV-header → canonical field ("" = don't import).
Response
Successful Response